SMILE Safety analysis and verification/validation of MachIne LEarning based systems
Purpose and goal
The purpose of SMILE is to explore the challenges while introducing machine learning-based systems in automated driving (AD) applications. It also aims to propose strategies to cope with those challenges to guide the industry and thus, be able to realize the potential to apply machine learning in safety critical systems.
Expected results and effects
SMILE resulted in a research agenda for a research program also named SMILE and an application for a continuation project. The continuation project aims at developing run-time monitoring for Deep Machine Learning (DML)-based perception using the concept of adaptive safety cage architectures a.k.a. safety supervisors, as a strategy to increasing the integrity towards faults when using DML in AD. An application for an Institute PhD student project within the area is also a result of the project. The project findings are published in two papers and two poster presentations.
Planned approach and implementation
The project was brought out in two parallel tracks, a theoretical state-of-art and a workshop series, and finalized with an analysis/synthesis. The theoretical framework of the industrial needs were found during the analysis formulated into a research agenda for the SMILE research program.